U.S. patent number 11,295,130 [Application Number 17/036,674] was granted by the patent office on 2022-04-05 for aerial line extraction system and aerial line extraction method.
This patent grant is currently assigned to HITACHI SOLUTIONS, LTD.. The grantee listed for this patent is HITACHI SOLUTIONS, LTD.. Invention is credited to Nobutaka Kimura, Kishiko Maruyama, Sadaki Nakano.
![](/patent/grant/11295130/US11295130-20220405-D00000.png)
![](/patent/grant/11295130/US11295130-20220405-D00001.png)
![](/patent/grant/11295130/US11295130-20220405-D00002.png)
![](/patent/grant/11295130/US11295130-20220405-D00003.png)
![](/patent/grant/11295130/US11295130-20220405-D00004.png)
![](/patent/grant/11295130/US11295130-20220405-D00005.png)
![](/patent/grant/11295130/US11295130-20220405-D00006.png)
![](/patent/grant/11295130/US11295130-20220405-D00007.png)
![](/patent/grant/11295130/US11295130-20220405-D00008.png)
![](/patent/grant/11295130/US11295130-20220405-D00009.png)
![](/patent/grant/11295130/US11295130-20220405-D00010.png)
View All Diagrams
United States Patent |
11,295,130 |
Maruyama , et al. |
April 5, 2022 |
Aerial line extraction system and aerial line extraction method
Abstract
Provided is an aerial line extraction system including: an
area-of-interest cropping unit that crops a region where a point
cloud data of aerial lines is assumed to exist as an area of
interest by setting coordinates of utility poles as a reference
from a three-dimensional point cloud data of a three-dimensional
shape that includes the aerial lines and trees installed in the air
via the utility poles; an aerial line candidate extraction unit
that extracts a candidate point cloud data of the aerial lines from
the three-dimensional point cloud data in the area of interest; and
an aerial line model estimation unit that estimates a model of the
aerial lines on the basis of the extracted candidate point cloud
data of the aerial lines.
Inventors: |
Maruyama; Kishiko (Tokyo,
JP), Kimura; Nobutaka (Tokyo, JP), Nakano;
Sadaki (Tokyo, JP) |
Applicant: |
Name |
City |
State |
Country |
Type |
HITACHI SOLUTIONS, LTD. |
Tokyo |
N/A |
JP |
|
|
Assignee: |
HITACHI SOLUTIONS, LTD. (Tokyo,
JP)
|
Family
ID: |
1000006221402 |
Appl.
No.: |
17/036,674 |
Filed: |
September 29, 2020 |
Prior Publication Data
|
|
|
|
Document
Identifier |
Publication Date |
|
US 20210103727 A1 |
Apr 8, 2021 |
|
Foreign Application Priority Data
|
|
|
|
|
Oct 7, 2019 [JP] |
|
|
JP2019-184232 |
|
Current U.S.
Class: |
1/1 |
Current CPC
Class: |
G06V
20/176 (20220101); G06V 10/30 (20220101); G06V
10/25 (20220101); G06K 9/6218 (20130101) |
Current International
Class: |
G06K
9/00 (20060101); G06K 9/62 (20060101) |
References Cited
[Referenced By]
U.S. Patent Documents
Foreign Patent Documents
|
|
|
|
|
|
|
2013126960 |
|
Jun 2013 |
|
JP |
|
2018-195240 |
|
Dec 2018 |
|
JP |
|
Other References
Jwa, Y., G. Sohn, and H. B. Kim. "Automatic 3d powerline
reconstruction using airborne lidar data." Int. Arch. Photogramm.
Remote Sens 38.Part 3 (2009): W8. (Year: 2009). cited by examiner
.
Zhu, Lingli, and Juha Hyyppa. "Fully-automated power line
extraction from airborne laser scanning point clouds in forest
areas." Remote Sensing 6.11 (2014): 11267-11282. (Year: 2014).
cited by examiner .
Clode, Simon, and Franz Rottensteiner. "Classification of trees and
powerlines from medium resolution airborne laserscanner data in
urban environments." Proceedings of the APRS Workshop on Digital
Image Computing (WDIC), Brisbane, Australia. vol. 21. 2005. (Year:
2005). cited by examiner.
|
Primary Examiner: Perlman; David
Attorney, Agent or Firm: Volpe Koenig
Claims
What is claimed is:
1. An aerial line extraction system comprising: a memory; an
input/output device; and a processor communicatively coupled to the
memory and the input/output device, wherein the processor is
configured to: crop a region where a point cloud data of aerial
lines is assumed to exist as an area of interest by setting
coordinates of utility poles as a reference from a
three-dimensional point cloud data of a three-dimensional shape
that includes the aerial lines and trees installed in the air via
the utility poles, extract a candidate point cloud data of the
aerial lines from the three-dimensional point cloud data in the
area of interest, estimate a model of the aerial lines on the basis
of the extracted candidate point cloud data of the aerial lines,
segment the three-dimensional point cloud data in the area of
interest by slice planes at regular intervals, generate a plurality
of clusters by clustering regions segmented by the slice planes by
distance as the region segmented by the slice plane, set a
threshold value a maximum distance determined, generate the
plurality of clusters using the threshold value, such that on a
condition that a point of the point cloud data is at a distance
that is less than or equal to the threshold value, the point is set
within a same cluster, and on a condition that the point is at a
distance that is greater than the threshold value, the point is set
within an other cluster, and extract the candidate point cloud data
of the aerial lines by classifying the plurality of clusters by a
predetermined size.
2. The aerial line extraction system according to claim 1, wherein
the processor segments the three-dimensional point cloud data in
the area of interest by the slice plane perpendicular to the ground
and perpendicular to a line connecting between the pair of utility
poles, generates the plurality of clusters by clustering a
rectangular parallelepiped region, and extracts the candidate point
cloud data of the aerial line by classifying the plurality of
clusters as the predetermined size by a minimum bounding box
size.
3. The aerial line extraction system according to claim 2, wherein
the processor determines whether or not the circumscribed rectangle
size falls within a predetermined dimensional range, in a case
where the minimum bounding box size falls within the predetermined
dimensional range as a result of the determination, determines the
point cloud data to be the candidate point cloud data of the aerial
line, and in a case where the minimum bounding box size does not
fall within the predetermined dimensional range as a result of the
determination, determines the point cloud data to be a point cloud
data other than the candidate point cloud data for the aerial
line.
4. The aerial line extraction system according to claim 3, wherein,
in a case where the minimum bounding box size does not fall within
the predetermined dimensional range as the result of the
determination, the processor determines the point cloud data other
than the candidate point cloud data for the aerial line to be the
point cloud data of the tree and removes the point cloud data of
the tree as noise from the three-dimensional point cloud data in
the area of interest.
5. The aerial line extraction system according to claim 1, further
comprising a display device that displays the model of the aerial
line estimated by the processor.
6. The aerial line extraction system according to claim 3, wherein
the predetermined dimensional range is defined by:
SegX.gtoreq.MinSegX, SegY<MaxSegY, SegZ<MaxSegZ, wherein SegX
is a difference between a maximum X coordinate and a minimum X
coordinate, MinSegX is the threshold value in the X axis, SegY is a
difference between a maximum Y coordinate and a minimum Y
coordinate, MaxSegY is the threshold value in the Y axis, SegZ is a
difference between a maximum Z coordinate and a minimum Z
coordinate, and MaxSegZ is the threshold value in the Z axis.
7. An aerial line extraction system, comprising: a memory; an
input/output device; and a processor communicatively coupled to the
memory and the input/output device, wherein the processor is
configured to: crop a region where a point cloud data of an aerial
line is assumed to exist as an area of interest by setting
coordinates of a utility pole as a reference from a
three-dimensional point cloud data of a three-dimensional shape
that includes the aerial line and the tree installed in the air via
a utility pole, extract a candidate point cloud data of the aerial
line from the three-dimensional point cloud data in the area of
interest, estimate a model of the aerial line on the basis of the
extracted candidate point cloud data of the aerial line, extract
the candidate point cloud data of the aerial line by removing the
point cloud data of the tree as noise from the three-dimensional
point cloud data in the area of interest by segmenting the
three-dimensional point cloud data in the area of interest by slice
planes at regular intervals, generating a plurality of clusters by
clustering regions segmented by the slice planes by distance as the
region segmented by the slice plane, and setting a threshold value
a maximum distance determined, generating the plurality of clusters
using the threshold value, such that on a condition that a point of
the point cloud data is at a distance that is less than or equal to
the threshold value, the point is set within a same cluster, and on
a condition that the point is at a distance that is greater than
the threshold value, the point is set within an other cluster.
8. An aerial line extracting method comprising: a three-dimensional
point cloud data acquiring step of acquiring a three-dimensional
point cloud data of a three-dimensional shape including aerial
lines and trees installed in the air via utility poles; an
area-of-interest cropping step of cropping a region where a point
cloud data of the aerial lines is assumed to exist as an area of
interest by setting coordinates of the utility poles as a reference
from a three-dimensional point cloud data; an aerial line candidate
extracting step of extracting a candidate point cloud data of the
aerial lines from the three-dimensional point cloud data in the
area of interest; and an aerial line model estimating step of
estimating a model of the aerial lines on the basis of the
extracted candidate point cloud data of the aerial lines, wherein
the aerial line candidate extracting step segments the
three-dimensional point cloud data in the area of interest by slice
planes at regular intervals; generates a plurality of clusters by
clustering regions segmented by the slice planes by distance as the
region segmented by the slice plane, sets a threshold value a
maximum distance determined, generates the plurality of clusters
using the threshold value, such that on a condition that a point of
the point cloud data is at a distance that is less than or equal to
the threshold value, the point is set within a same cluster, and on
a condition that the point is at a distance that is greater than
the threshold value, the point is set within an other cluster, and
extracts the candidate point cloud data of the aerial lines by
classifying the plurality of clusters by a predetermined size.
9. The aerial line extracting method according to claim 8, wherein
the aerial line candidate extracting step segments the
three-dimensional point cloud data in the area of interest by the
slice planes that are perpendicular to the ground and perpendicular
to a line connecting between a pair of the utility poles, generates
a plurality of the clusters by clustering a rectangular
parallelepiped region, and extracts the candidate point cloud data
of the aerial line by classifying a plurality of the clusters with
a circumscribed rectangle size as the predetermined size.
10. The aerial line extracting method according to claim 9, wherein
the aerial line candidate extracting step determines whether or not
the minimum bounding box size falls within a predetermined
dimensional range, in a case where the minimum bounding box size
falls within the predetermined dimensional range as a result of the
determination, determines the point cloud data to be the candidate
point cloud data of the aerial line, and in a case where the
minimum bounding box size does not fall within the predetermined
dimensional range as a result of the determination, determines the
point cloud data to be a point cloud data other than the candidate
point cloud data for the aerial line.
11. The aerial line extracting method according to claim 10,
wherein, in a case where the minimum bounding box size does not
fall within the predetermined dimensional range as the result of
the determination, the aerial line candidate extracting step
determines the point cloud data other than the candidate point
cloud data for the aerial line to be the point cloud data of the
tree and removes the point cloud data of the tree as noise from the
three-dimensional point cloud data in the area of interest.
12. The aerial line extracting method according to claim 8, further
comprising a display step of displaying a model of the aerial line
estimated in the aerial line model estimating step.
Description
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to an aerial line extraction system
and an aerial line extracting method.
2. Description of the Related Art
There is known a mobile mapping system (MMS) in which an inspection
vehicle is equipped with measuring devices such as a
three-dimensional laser scanner (laser distance measuring device),
a digital camera, and a GPS, and while traveling, the inspection
vehicle collects a three-dimensional shape of a terrain and
structures around a road in a form of three-dimensional point cloud
data.
The MMS can efficiently and accurately acquire a wide range of
three-dimensional point cloud data around the road, and thus, the
MMS has been expected to be used for checking the situations of
facilities such as utility poles and aerial lines (electric lines
and communication lines) around the road.
For example, in JP 2018-195240 A, three-dimensional model data of
facilities is generated on the basis of three-dimensional point
cloud data acquired by the MMS, and the thickness, inclination
angle, and deflection of the utility pole (pole) and the tree and
the minimum ground height of the aerial line (cable) are calculated
on the basis of this three-dimensional model data. In addition, the
three-dimensional model data are superimposed on the image data
imaged by the digital camera so as to match the respective position
coordinates, and the parameter information indicating the
structures of the utility poles, the trees and the aerial lines is
indicated on the three-dimensional superimposed image.
JP 2018-195240 A discloses a method of generating three-dimensional
model data of utility poles, trees, and aerial lines on the basis
of three-dimensional point cloud data and superimposing the
three-dimensional model data and image data to visualize the data
for the purpose of detecting outdoor facilities such as utility
poles and aerial lines.
In JP 2018-195240 A, a utility pole and a tree are modeled as a
three-dimensional object in which circles are vertically
overlapped, and an aerial line is modeled as a three-dimensional
object in which catenary curves are connected. In a process of
detecting the aerial line, noise removal is performed by removing
unnatural catenary curves. However, in a case where trees become
noise in mountainous areas, a large number of catenary curves
become candidates for the aerial lines, it is difficult to remove
the noise.
SUMMARY OF THE INVENTION
An object of the present invention is to enable extraction of
aerial lines even in a case where trees become noise by separating
aerial lines and noises such as trees from three-dimensional point
cloud data in an aerial line extraction system, and after that,
performing model estimation.
According to an aspect of the present invention, there is provided
an aerial line extraction system including: an area-of-interest
cropping unit that crops a region where a point cloud data of
aerial lines is assumed to exist as an area of interest by setting
coordinates of utility poles as a reference from a
three-dimensional point cloud data of a three-dimensional shape
that includes the aerial lines and trees installed in the air via
the utility poles; an aerial line candidate extraction unit that
extracts a candidate point cloud data of the aerial lines from the
three-dimensional point cloud data in the area of interest; and an
aerial line model estimation unit that estimates a model of the
aerial lines on the basis of the extracted candidate point cloud
data of the aerial lines, in which the aerial line candidate
extraction unit segments the three-dimensional point cloud data in
the area of interest by slice planes at regular intervals,
generates a plurality of clusters by clustering regions segmented
by the slice planes, and extracts the candidate point cloud data of
the aerial lines by classifying the plurality of clusters by a
predetermined size.
According to another aspect of the present invention, there is
provided an aerial line extraction system, including: an
area-of-interest cropping unit that crops a region where a point
cloud data of an aerial line is assumed to exist as an area of
interest by setting coordinates of a utility pole as a reference
from a three-dimensional point cloud data of a three-dimensional
shape that includes the aerial line and the tree installed in the
air via a utility pole; an aerial line candidate extraction unit
that extracts a candidate point cloud data of the aerial line from
the three-dimensional point cloud data in the area of interest; and
an aerial line model estimation unit that estimates a model of the
aerial line on the basis of the extracted candidate point cloud
data of the aerial line, in which the aerial line candidate
extraction unit extracts the candidate point cloud data of the
aerial line by removing the point cloud data of the tree as noise
from the three-dimensional point cloud data in the area of
interest.
According to still another aspect of the present invention, there
is provided an aerial line extracting method including: a
three-dimensional point cloud data acquiring step of acquiring a
three-dimensional point cloud data of a three-dimensional shape
including aerial lines and trees installed in the air via utility
poles; an area-of-interest cropping step of cropping a region where
a point cloud data of the aerial lines is assumed to exist as an
area of interest by setting coordinates of the utility poles as a
reference from a three-dimensional point cloud data; an aerial line
candidate extracting step of extracting a candidate point cloud
data of the aerial lines from the three-dimensional point cloud
data in the area of interest; and an aerial line model estimating
step of estimating a model of the aerial lines on the basis of the
extracted candidate point cloud data of the aerial lines, in which
the aerial line candidate extracting step segments the
three-dimensional point cloud data in the area of interest by slice
planes at regular intervals; generates a plurality of clusters by
clustering regions segmented by the slice planes, and extracts the
candidate point cloud data of the aerial lines by classifying the
plurality of clusters by a predetermined size.
According to one aspect of the present invention, in an aerial line
extraction system, by separating aerial lines and noise such as
trees from three-dimensional point cloud data, and after that, by
performing model estimation, the aerial lines can be extracted even
in a case where the trees become noise.
BRIEF DESCRIPTION OF THE DRAWINGS
FIG. 1 is a perspective view illustrating an example in which a
three-dimensional point cloud data is acquired by a laser distance
measuring device.
FIG. 2 is a conceptual diagram illustrating the three-dimensional
point cloud data acquired by the laser distance measuring
device.
FIG. 3 is a block diagram illustrating a configuration of an aerial
line extraction system according to a first embodiment.
FIG. 4 is a flowchart illustrating a processing flow of the aerial
line extraction system according to the first embodiment.
FIG. 5A is a diagram illustrating a setting example of an area of
interest and a local coordinate system and is a schematic diagram
of the three-dimensional point cloud data of FIG. 2 viewed from the
above.
FIG. 5B is a diagram illustrating a setting example of an area of
interest and a local coordinate system, and is a schematic diagram
of the three-dimensional point cloud data of FIG. 2 viewed from the
side.
FIG. 6 is a flowchart illustrating a processing flow of aerial line
candidate point cloud extraction.
FIG. 7A is a diagram describing an example of slicing an area of
interest perpendicularly to the ground.
FIG. 7B is a diagram describing an example of slicing an area of
interest perpendicularly to the ground.
FIG. 8A is a diagram illustrating an example of an effect based on
the first embodiment and illustrates a point cloud as a result of
aerial line candidate extraction by slice segmentation.
FIG. 8B is a diagram illustrating an example of an effect based on
the first embodiment and illustrates a result of aerial line model
estimation.
FIG. 9A is a diagram illustrating a result of a case where first
embodiment is not applied and illustrates a point cloud in which an
area of interest is cropped.
FIG. 9B is a diagram illustrating a result of a case where first
embodiment is not applied and illustrates a result of aerial line
model estimation.
FIG. 10 is a conceptual diagram illustrating a display example of a
three-dimensional point cloud data including a lead-in line
according to a second embodiment.
FIG. 11 is a conceptual diagram of a three-dimensional point cloud
including the lead-in line according to the second embodiment as
viewed down from the above.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
Hereinafter, embodiments will be described with reference to the
drawings.
First Embodiment
An example in which a three-dimensional point cloud data is
acquired by a laser distance measuring device will be described
with reference to FIG. 1. For example, the laser distance measuring
device 102 is mounted on a vehicle 101, and the vehicle is allowed
to travel while measuring the periphery of the vehicle 101. The
laser distance measuring device 102 is a device that scans a laser
beam 103 at predetermined intervals and collects three-dimensional
point cloud data of the surrounding in a three-dimensional shape. A
wide range of a three-dimensional map data can be generated by
integrating the three-dimensional point cloud data collected while
traveling. Such a technique is known as Mobile Mapping System
(MMS).
The three-dimensional point cloud data includes not only data of
roads 104, utility poles 105, or buildings such as buildings and
signs but also includes data of aerial lines 106 such as electric
lines or communication lines installed through the utility poles
105 in the air and trees 107. In addition, although not illustrated
in FIG. 1, the aerial line 106 includes a lead-in line from the
utility pole 105 to each house, a branch line supporting the
utility pole, and the like. Such arrangement information of the
aerial lines 106 is useful for the time of performing maintenance
of electric lines, communication lines, and the like.
An example of the three-dimensional point cloud data acquired in
FIG. 1 will be described with reference to FIG. 2. The
three-dimensional point cloud data includes a point group 204 of
roads, a point group 205 of utility poles, a point cloud 206 of
aerial lines, and a point cloud 207 of trees.
The configuration of the aerial line extraction system according to
the first embodiment will be described with reference to FIG. 3.
The aerial line extraction system according to the first embodiment
includes a processing device 300. The processing device 300
includes an area-of-interest cropping unit 301, an aerial line
candidate extraction unit 302, and an aerial line model estimation
unit 303. In addition, a three-dimensional point cloud data file
304, a facility database (facility DB) 305, an aerial line model
file 306, a display 307 as a display device, and a keyboard/mouse
308 as an input device are connected to the processing device 300.
Herein, the processing device 300 is configured with, for example,
a processor (CPU, GPU, or the like). In addition, the
area-of-interest cropping unit 301, the aerial line candidate
extraction unit 302, and the aerial line model estimation unit 303
are configured with, for example, a processor (CPU, GPU, or the
like).
The three-dimensional point cloud data file 304 stores the
three-dimensional point cloud data acquired by the laser distance
measuring device 102. As illustrated in FIG. 2, the
three-dimensional point cloud data includes a point cloud data 204
of the road 104, a point cloud data 205 of the utility pole 105, a
point cloud data 206 of the aerial line 106, and a point cloud data
207 of the tree 107.
The facility DB 305 stores a management data of the utility pole
105 registered in advance. The management data of the utility pole
105 includes information of, for example, a management number, a
type (model number, diameter, height), an installation position
(address, coordinates), and the like.
The area-of-interest cropping unit 301 sets a region where the
point cloud 206 of the aerial line 106 is assumed to exist as an
area of interest by setting coordinates of the utility pole 105
stored in the facility DB 305 as a reference from the
three-dimensional point cloud data stored in the three-dimensional
point cloud data file 304 and extracts a point cloud data in the
area of interest.
The aerial line candidate extraction unit 302 removes noise such as
the trees 107 from the three-dimensional point cloud data in the
area of interest and extracts the point cloud data that is a
candidate for the aerial line 106.
The aerial line model estimation unit 303 estimates an aerial line
model represented by a catenary curve (or a quadratic curve
approximation of the catenary curve) on the basis of the
three-dimensional point cloud data that is a candidate for the
aerial line 106 and outputs the parameter of the aerial line model
to the aerial line model file 306. The model estimation performed
by the aerial line model estimation unit 303 is based on a general
model estimation method such as Random Sample Consensus
(RANSAC).
The display 307 is a display device that displays the
three-dimensional point cloud data and the aerial line model and
displays, for example, the three-dimensional point cloud data and
the aerial line model as illustrated in FIGS. 8A, 8B, 9A, and 9B.
The keyboard/mouse 308 is an example of an input device. The input
device specifies the utility pole 105 and the like in the
area-of-interest cropping unit 301.
The processing flow of the aerial line extraction system according
to the first embodiment will be described with reference to FIG. 4.
The aerial line extraction system is realized by allowing a general
information processing apparatus to process software.
In step S401, the three-dimensional point cloud data acquired by
the method described in FIG. 1 is input from the three-dimensional
point cloud data file 304. The three-dimensional point cloud data
is a data of a point cloud as illustrated in FIG. 2. For example,
in a case where a perpendicular coordinate system is employed, each
point is represented by coordinates (x, y, z). The origin (0, 0, 0)
of the coordinates and the x-axis, y-axis, and z-axis can be set
arbitrarily, but in general, the (x, y) coordinates are defined in
a form that can be associated with a map, and in many cases, the
z-axis is taken to be in the vertical direction. Herein, the
coordinate system of the input three-dimensional point cloud data
is referred to as a world coordinate system, and the coordinate
system that is subjected to coordinate conversion for easy
processing is referred to as a local coordinate system. In
addition, a spherical coordinate system or another coordinate
system may be used instead of the perpendicular coordinate
system.
In step S402, the area-of-interest cropping unit 301 crops a
portion where the point cloud 206 of the aerial line is likely to
exist as an area of interest from the three-dimensional point cloud
data. The cropping method is not particularly limited, but for
example, as illustrated in FIGS. 5A and 5B, the area of interest
501 is defined and cropped on the basis of the coordinates of two
utility poles.
The area-of-interest cropping unit 301 obtains the coordinates that
define the area of interest 501 by setting the utility pole
coordinates and the like as a reference. After that, the
area-of-interest cropping unit 301 extracts the point cloud in the
area of interest 501 from the three-dimensional point cloud data
stored in the three-dimensional point cloud data file 304.
The concept of the extraction processing of the area of interest in
step S402 will be described with reference to FIGS. 5A and 5B.
Herein, FIG. 5A is a schematic view of the three-dimensional point
cloud data of FIG. 2 viewed from the above, and FIG. 5B is a
schematic view of the three-dimensional point cloud data of FIG. 2
viewed from the side. For the explanation, a local coordinate
system is employed, and the longitudinal direction (extension
direction) of the point cloud 206 of the aerial line is set as x,
the height direction (direction perpendicular to the ground) is set
as z, and the direction perpendicular to x and z is set as y.
As the area of interest 501, for example, a rectangular
parallelepiped region parallel to a line connecting a pair of
utility pole coordinates (x.sub.1, y.sub.1) and (x.sub.2, y.sub.2)
is set. In the local coordinate system 502, for example, the
midpoint between a pair of utility pole coordinates is set as the
origin, the direction connecting the pair of utility pole
coordinates is set as the x-axis, the direction parallel to the
ground and perpendicular to the x-axis is set as the y axis, and
the direction perpendicular to the ground is set as the z-axis.
As a method of designating a pair of utility pole coordinates
(x.sub.1, y.sub.1) and (x.sub.2, y.sub.2), for example, there is a
method in which a user selects two utility poles 105 by using the
keyboard/mouse 308 on a screen of the display 307 displaying a data
list of the utility poles 105 searched from the facility DB 305.
Alternatively, there is a method in which a user selects two
utility poles 105 by using the keyboard/mouse 308 on the screen of
the display 307 displaying the installation position of the utility
poles 105 on a map. Furthermore, there is a method of selecting a
combination of two neighboring utility poles 105 from the
coordinate information of the utility pole 105 registered in the
facility DB 305 in an indiscriminative manner.
As illustrated in FIGS. 5A and 5B, in the area of interest 501
defined by the pair of utility pole coordinates, point clouds such
as the point cloud 204 of roads and the point cloud 205 of utility
poles are highly likely to be removed as outside of the area of
interest 501, and the point cloud 206 of aerial lines or the point
cloud 207 of trees are highly likely to remain as inside of the
area of interest 501.
In step S403, after cropping the area of interest 501, the world
coordinate system is converted to the local coordinate system in
which two utility poles 105 are on the x-axis, and the centers of
the two utility poles 105 become origins (x, y)=(0, 0).
In addition, the area-of-interest extraction in step S402 and the
conversion to the local coordinate system in step S403 are not
limited to this order, and the order of processing may be
changed.
In step S404, the aerial line candidate extraction unit 302
performs the extraction of the point cloud of aerial line
candidates. The aerial line candidate extraction unit 302 extracts
a candidate point cloud data of the aerial line 106 from the
three-dimensional point cloud data in the area of interest 501.
In step S405, estimation of the aerial line model is performed. The
aerial line model estimation unit 303 estimates the model of the
aerial line 106 on the basis of the extracted candidate point cloud
data of the aerial line 106. In step S406, conversion to the world
coordinate system is performed. In step S406, in a case where the
local coordinate system has been converted in step S403, the
coordinate system is returned to the world coordinate system. In
step S407, the outputting of the aerial line model is performed.
Specifically, the model of the aerial line 106 estimated by the
aerial line model estimation unit 303 is output to the aerial line
model file 306. In addition, if necessary, the aerial line model of
the estimation result is superimposed and displayed on the display
307 which displays the point cloud or the like of the aerial line
candidates.
With reference to FIG. 6, the processing flow of the he aerial line
candidate point cloud extraction in step S404 of FIG. 4 will be
described.
In step S601, the area of interest 501 is sliced perpendicularly to
the ground.
An example of slicing the area of interest 501 perpendicularly to
the ground will be described with reference to FIGS. 7A and 7B.
In step S601, the area of interest 501 cropped from the
three-dimensional point cloud data as a region where the point
cloud 206 of the aerial lines is likely to exist is sliced by the
slice planes 701 that are perpendicular to the ground and
perpendicular to a line connecting the point clouds 205 of a pair
of utility poles. In this manner, the three-dimensional point cloud
data in the area of interest 501 is segmented by the slice planes
701 at regular intervals. That is, the three-dimensional point
cloud data in the area of interest 501 is segmented by the slice
planes 701 that are perpendicular to the ground and perpendicular
to the line connecting the point clouds 205 of the pair of utility
poles. The rectangular parallelepiped regions (simply, referred to
as slices) segmented by the slice planes 701 usually have the same
shape and the same volume.
In step S602, each slice is subjected to clustering by
distance.
The clustering is based on a general clustering method. For
example, the maximum value of the distance determined to be the
same cluster is set as a threshold value parameter, and in a case
where the distance to the nearest point is equal to or less than
the threshold value, the cluster is determined to be the same
cluster.
In step S603, each cluster is classified by the minimum bounding
box size (step S6031). Specifically, it is determined whether or
not the following mathematical Formula 1 is satisfied. In
mathematical Formula 1, SegX, SegY, and SegZ represent minimum
bounding box sizes. SegX represents a difference between the
maximum X coordinate and the minimum X coordinate of the point
group in the cluster, SegY represents a difference between the
maximum Y coordinate and the minimum Y coordinate, and SegZ
represents a difference between the maximum Z coordinate and the
minimum Z coordinate. MinSegX, MaxSegY, and MaxSegZ are threshold
value parameters set in the aerial line extraction system.
SegX.gtoreq.MinSegX, SegY<MaxSegY, SegZ<MaxSegZ [mathematical
Formula 1]
In a case where the result of the determination is that the
above-described mathematical Formula 1 is satisfied, the point
cloud is set as a candidate for the point cloud 206 of aerial lines
(step S6032). In a case where the result of the determination is
that the above-described mathematical Formula 1 is not satisfied,
the point group is determined as a point cloud candidate other than
the candidate of the point cloud 206 of aerial lines (step
S6033).
As described above, in the processing of the aerial line candidate
point cloud extraction in step S404 of FIG. 4, the aerial line
candidate extraction unit 302 segments the three-dimensional point
cloud data in the area of interest 501 by the slice planes 701 at
regular intervals, and the areas segmented by the slice planes 701
are clustered to generate a plurality of clusters. Then, the
plurality of clusters are classified by a predetermined size, and
thus, the candidate point cloud data of the aerial line 106 is
extracted.
Specifically, the aerial line candidate extraction unit 302
segments the three-dimensional point cloud data in the area of
interest 501 by the slice plane 701 that is perpendicular to the
ground and perpendicular to a line connecting the point clouds 205
of the pair of utility poles and generates a plurality of clusters
by clustering the rectangular parallelepiped regions segmented by
the slice plane 701 according to the distance. Then, the plurality
of clusters are classified with the minimum bounding box size, and
thus, the candidate point cloud data of the aerial line 106 is
extracted.
In addition, the aerial line candidate extraction unit 302
determines whether or not the minimum bounding box size falls
within a predetermined dimensional range (dimensional range defined
by the above-described mathematical Formula 1). In a case where the
result of the determination is that the minimum bounding box size
falls within the predetermined dimensional range (in a case where
the above-described mathematical Formula 1 is satisfied), the point
cloud data is determined to be a candidate point cloud data of the
aerial line 106. In a case where the result of the determination is
that the minimum bounding box size does not fall within the
predetermined dimensional range (in a case where the
above-described mathematical Formula 1 is not satisfied), the point
cloud data is determined to be a point cloud data other than the
candidate point cloud data of the aerial line 106. For example, in
a case where the result of the determination is that the minimum
bounding box size does not fall within the predetermined
dimensional range (in a case where the above-described mathematical
Formula 1 is not satisfied), the point cloud data other than the
candidate point cloud data of the aerial line 106 is determined as
the point cloud data of the tree 107, and thus, the point cloud
data of the tree 107 is removed as noise from the three-dimensional
point cloud data in the area of interest 501.
In this manner, in the above-described aerial line candidate point
cloud extraction processing, by performing the slicing processing
(step S601) and the clustering processing (step S602) illustrated
in FIG. 6, the point cloud 207 of the trees and the point cloud 206
of the aerial lines can be separated highly accurately.
FIGS. 8A and 8B are conceptual diagrams illustrating an example of
the effect based on the first embodiment, FIG. 8A illustrates a
point cloud of aerial line candidate extraction results by slice
segmentation, and FIG. 8B illustrates a result of aerial line model
estimation.
In FIG. 8A, 8B indicates a point cloud of aerial line candidates in
the area of interest 501. The extraction of the point cloud 801 of
aerial line candidates in the area of interest 501 is performed by
the aerial line candidate extraction unit 302 in FIG. 3 through
step S404 in FIG. 4. As illustrated in FIG. 8A, the point cloud 801
of aerial line candidates is extracted in the area of interest 501,
and noise of the point cloud 207 of trees or the like is
removed.
In FIG. 8B, 802 indicates the result of the aerial line model
estimation estimated on the basis of the point cloud 801 of aerial
line candidates in the area of interest 501. The aerial line model
estimation is performed by the aerial line model estimation unit
303 in FIG. 3 through step S405 in FIG. 4. As illustrated in FIG.
8B, the aerial line model 802 corresponding to the point cloud 801
of aerial line candidates is estimated.
FIGS. 9A and 9B are conceptual diagrams illustrating results of a
case where the first embodiment is not applied, FIG. 9A illustrates
a point cloud in which the area of interest is cropped, and FIG. 9B
illustrates a result of the aerial line model estimation.
As illustrated in FIG. 9A, in a point cloud 901 of the area of
interest 501, a point cloud 207 of trees and a point cloud 206 of
aerial lines are mixed.
As a result, in a result 902 of the aerial line model estimation
illustrated in FIG. 9B, the point cloud 207 of trees and the point
cloud 206 of aerial lines are not separated, and both the point
cloud 207 of trees and the point cloud 206 of aerial lines become
aerial line candidates. For this reason, an infinite number of
aerial line models 902 are estimated. This is because the
extraction of the point cloud 801 of aerial line candidates in the
area of interest 501 is not performed by the aerial line candidate
extraction unit 302 in FIG. 3 as in FIGS. 8A and 8B.
According to the first embodiment, the aerial line candidate
extraction unit 302 removes the point cloud data of the trees 107
from the three-dimensional point cloud data in the area of interest
501 as noise, and thus, extracts the candidate point cloud data of
the aerial lines 106. As a result, the point cloud 207 of trees and
the point cloud 206 of aerial lines can be separated, and
particularly, the aerial lines 106 can be extracted even in a case
where the trees 107 become noise in a mountainous area.
Second Embodiment
In a second embodiment, an example of a lead-in line and a branch
line as an example of an aerial line will be described. The lead-in
line is a cable for wiring between the utility pole and each house
and usually denotes a line from the utility pole to the
lead-in-line attachment point attached to the eaves of each house.
A branch line is a wire for supporting the utility pole. Most of
the system configuration and the processing flow may be configured
similarly to the first embodiment. Hereinafter, the portions
different from those of the first embodiment will be described.
FIG. 10 is a display example of the three-dimensional point cloud
data including a lead-in line and a branch line.
Reference numeral 1001 indicates a point cloud of a lead-in line
(connecting a utility pole and each house). Reference numeral 1002
indicates a point cloud of a branch line (installed to support the
utility pole 205). Reference numeral 1003 indicates a point cloud
of a building. The point cloud 1001 of the lead-in line from the
utility pole 205 is connected to a lead-in-line attachment point of
the point cloud 1003 of a building. In addition, point cloud data
such as the point cloud 207 of trees is included. In the case of
setting the point cloud 1001 of such lead-in lines as a target, the
case can be dealt with by changing the an area-of-interest cropping
processing step S402 in FIG. 4 and the an area of interest slicing
processing step S601 in FIG. 6 in the processing according to the
first embodiment illustrated in the flow of FIG. 4.
FIG. 11 is a conceptual diagram of the three-dimensional point
cloud including the point cloud 1001 of the lead-in line and the
point cloud 1002 of branch line of FIG. 10. As illustrated in FIG.
11, the lead-in line and the branch line are lines extending
radially from the utility pole as a base point.
The slice plane 1101 is a concentric cylinder centered on the point
cloud 205 of the utility pole. The point cloud 1001 of lead-in
lines is led from the point cloud 205 of utility poles to the point
cloud 1003 of buildings. In the second embodiment, in the
area-of-interest cropping processing (S402) illustrated in FIG. 4,
a cylinder centered on the point cloud 205 of the utility poles is
defined as the area of interest by designating the point cloud 205
of the utility poles.
In the area of interest slicing processing step S601 illustrated in
FIG. 6, by slicing the area of interest by concentric cylinders,
donut-shaped subdivided areas are obtained. Specifically, the area
of interest is segmented by cylinder planes of concentric circles
that are perpendicular to the ground and are equally spaced in the
radial direction with the utility pole as the central axis. The
other processing is almost the same as that according to the first
embodiment, and the description thereof is omitted.
According to the second embodiment, the aerial line can be
extracted even in a case where the lead-in line or the branch line
is included.
* * * * *